OpenAI just published the map of how 300 million people are actually using AI at work, and the data shows we're already past the "will AI change work" debate.
The Summary
- OpenAI released Signals data tracking 165 work activities across nearly 100 countries, showing real-world ChatGPT usage patterns at scale
- The data reveals which work tasks are being automated first and how usage evolves as people get comfortable with AI assistance
- This is the largest public dataset on how AI tools are actually being integrated into daily work, not just speculated about in think pieces
The Signal
OpenAI's Signals dataset maps 165 distinct work activities across nearly 100 countries, tracking how 300 million ChatGPT users are deploying AI in their actual jobs. This isn't survey data or market research. It's behavioral data from people who are already working differently.
The methodology matters here. OpenAI is tracking specific work tasks, from writing code to drafting legal documents to analyzing financial data. The granularity means we can see which activities are being augmented first and which ones people still handle manually. Geography matters too: usage patterns vary significantly by country, suggesting cultural and regulatory factors shape AI adoption as much as technology access does.
"This is the first comprehensive view of the agent economy in motion, captured from actual usage patterns rather than forecasts."
The timing of this release is strategic. OpenAI is building legitimacy for AI at work by showing it's already happening at scale. But the more important signal is what the data reveals about velocity. Usage patterns are changing over time, meaning people who started with basic tasks are moving to more complex work. The learning curve is real, but it's also fast.
Key patterns emerging from the data:
- Creative and analytical work is being augmented more than rote tasks
- Usage intensity correlates with knowledge work density, not just GDP
- Adoption curves vary wildly by country, suggesting regulatory and cultural friction points
This dataset will become a benchmark. Enterprises evaluating AI deployment now have real-world usage patterns to reference. Investors funding AI infrastructure can see which work categories are seeing the highest adoption velocity. Workers can see which skills are being augmented versus replaced.
The Implication
The next phase isn't about whether AI changes work. It's about the gap between early adopters and everyone else. Companies that use this data to understand where their employees are on the adoption curve can accelerate training and tool integration. Workers who study which activities are being automated fastest can position themselves accordingly.
Watch for two things: how enterprise software companies use this data to prioritize which workflows to rebuild, and how governments use it to shape workforce policy. The map is drawn. Now everyone's deciding where they fit on it.